Re: enhancement/384: mfactors (fwd)

From: Shalom.Bresticker@freescale.com
Date: Tue May 11 2004 - 01:50:00 PDT

  • Next message: Shalom Bresticker: "errata/583: comment_text allows \n and */"

    The following reply was made to PR enhancement/384; it has been noted by GNATS.

    From: Shalom.Bresticker@freescale.com
    To: btf-bugs@boyd.com
    Cc:
    Subject: Re: enhancement/384: mfactors (fwd)
    Date: Tue, 11 May 2004 12:05:50 +0300 (IDT)

     ---------- Forwarded message ----------
     Date: Mon, 10 May 2004 14:01:41 -0400
     From: Geoffrey.Coram <Geoffrey.Coram@analog.com>
     To: Shalom Bresticker <Shalom.Bresticker@motorola.com>
     Subject: Re: mfactors
     
     Shalom -
     In analog design, it is often desirable to put down
     several "identical" instances rather than one single one.
     
     The idea is to use a "weak law of large numbers" approach
     that means that the effective resistance of a string of
     several random resistors will have a smaller variance
     (by 1/sqrt(m)) than a single resistor -- assuming that
     the randomness of the resistor is independent of that
     of its neighbors.
     
     And in fact, for differential signals, the "common mode"
     randomness is eliminated by the design (if all the
     resistors are 10% below nominal, then the differential
     gain of the system is unchanged). So it is the local
     randomization, due to different etch rates or differences
     in the local topology (this resistor is next to a
     transistor, so the poly doesn't grow quite the same as
     the other one).
     
     The m-factor provides an easy way to say: give me 10
     identical resistors, rather than forcing the designer
     to cut&paste 9 times.
     
     -Geoffrey
     

    This archive was generated by hypermail 2.1.4 : Tue May 11 2004 - 01:50:04 PDT and
    sponsored by Boyd Technology, Inc.